Updated daily at 05:00 UTC
OCR document
DeepSeek-OCR
Document OCR models that compress page images into few vision tokens.
By DeepSeek.
Versions
2 versions
| Model | Maker | Kind | Released | Context | Input $/1M | Output $/1M |
|---|---|---|---|---|---|---|
| DeepSeek-OCR 2DeepSeek-OCR · Open weights | DeepSeek | OCR document | 27 Jan 2026 | — | — | — |
| DeepSeek-OCRDeepSeek-OCR · Legacy · Open weights | DeepSeek | OCR document | Oct 2025 | — | — | — |
